1

Australasian Vascular Audit Public Report – 2019

2

Contents Page

Foreword A Hill (President ANZSVS) 3

Introduction 4

Audit monitoring committee 4

Overview 5

Aortic surgery 13

i. Open aortic surgery 14

ii. Open Abdominal aortic aneurysms 17

ii. Endoluminal grafts (ELG) 19

iiii. Fenestrated and branched ELG 21

iv. Thoracic and thoracoabdominal 23

Carotid surgery 24

i. Carotid endarterectomy 24

ii. Carotid stents 28

Infrainguinal bypasses 29

i. Occlusion 32

ii. Amputation 33

Arterio-venous fistulae 34

Data validation and conclusions 36

Appendix 1-Algorithm for the outlier 38

Appendix 2-Statistical methods 39

Appendix 3-Features of the AVA 41

References 42

3

Foreword

It is with great pleasure that I present this AVA report. The AVA remains strong and is one of the

cornerstones of our Society.

The Healthcare Quality Improvement Partnership (HQIP) defines audit as “a quality improvement

process that seeks to improve patient care and outcomes through systematic review of care against

explicit criteria and the implementation of change”. I believe that we have built a strong platform to

achieve these goals while still looking to build further. The identification of adverse events is

important as well as the safe-guarding of our data. The Audit Monitoring committee independently

oversees these functions. Refinements to allow easier data entry and the ability to use handheld and

mobile devices is underway.

Our Data Administrator, Barry Beiles, provides us with a wealth of expertise, skill and dedication,

whilst overseeing its evolution.

A Hill

President ANZSVS

4

Introduction

The Australasian Vascular Audit (AVA) has just completed its 10th year of data collection. It was established in 2008 after constitutional changes had been adopted following a ballot with an overwhelming majority by the membership of the Australian and New Zealand Society for Vascular Surgery (ANZSVS). This had been a long-term goal of the Society with the aim of amalgamating the existing vascular audits throughout Australia and New Zealand. The audit is compulsory, with membership of the ANZSVS conditional upon participation in audit. Both public hospital and private practice data are collected at 2 points in the admission episode; at admission/operation and after discharge and only patients undergoing a surgical or endovascular procedure are entered in the database. Although all procedures are captured in the database, the following index procedures were selected for audit:

1. Aortic surgery –includes both aneurysmal and occlusive disease (survival)

i. Open elective and emergency

ii. Non-fenestrated elective and emergency endografts

iii. Fenestrated endografts

2. Carotid procedures (freedom from stroke/death)

i. Open carotid endarterectomy

ii. Carotid stents

3. Infrainguinal bypasses (patency and limb salvage)

4. AV Fistula for dialysis (patency)

Audit monitoring committee

The executive committee of the ANZSVS has established an Audit Monitoring Committee (AMC), which consists of 4 members; the Chairman of the AMC, the immediate past-president of the ANZSVS, the administrator of the AVA (a vascular surgeon with computer and statistical skills) and the president or immediate past-president of the Vascular Society of New Zealand (VSNZ). These members are elected and are senior members of the ANZSVS engaged in active vascular surgical practice. Their roles and responsibilities are:

▪ to oversee protection of the collected data ▪ to ensure confidentiality of participants (both surgeon and patient alike) ▪ to monitor the collection of the audit data and to facilitate maximal

compliance ▪ to prevent misuse of the data (including addressing complaints about misuse

of the data) ▪ to investigate and verify statistical outliers according to a pre-determined

algorithm ▪ to assess applications to determine suitability for participation in the AVA.

5

▪ to assess applications to use the collected data for non- audit purposes. ▪ to oversee the AVA verification process ▪ to provide an annual report of the AVA results for the ANZSVS AGM. ▪ to identify opportunities for performance improvement ▪ to identify opportunities for external publication ▪ to provide annual certificates of satisfactory vascular surgical audit

participation ▪ to oversee the disclosure of audit data to a third party at the instigation of a

participating member

Overview There were 43,387 operations entered in 2019; 38,521 from Australia and 4,866 from New Zealand (Fig 1). Although the demographic data applies to all operations, the outcome analyses are based on the 46,578 discharged patients (98.7%).

Fig 1. Volume of vascular surgery by country 2019

6

Fig 2. Operations by Australian State and New Zealand Region 2019

246 consultants entered data from 201 hospitals/clinics which are shown alphabetically in the following table.

Albany Day Hospital-Mira Mar

Alfred Hospital-Melbourne

Allamanda Private Hospital-Southport

Armadale Kelmscott District Hospital-Armadale

Ascot Hospital-Remuera

Ashford Hospital-Ashford

Auburn Hospital-Auburn

Auckland City Hospital-Auckland

Austin Hospital-Heidelberg

xxx rooms-QLD

Ballarat Base Hospital-North Ballarat

Ballina District Hospital-Ballina

Bankstown Hospital-Bankstown

Baringa Private Hospital-Coff's Harbour

Bentley Health Service-Bentley

Blacktown Hospital-Blacktown

Blue Mountains Hospital-Katoomba

Box Hill Hospital-Box Hill

7

Brisbane Waters Private Hospital-Woy Woy

Buderim Private Hospital-Buderim

Cabrini Hospital-Brighton

Cabrini Hospital-Malvern

Cairns Base Hospital-Cairns

Cairns Private Hospital-Cairns

Calvary Hospital-Lenah Valley

Calvary Hospital-North Adelaide

Calvary John James Hospital-Deakin

Calvary Private Hospital-Bruce

Calvary Public Hospital-Bruce

Calvary Wakefield Hospital-Adelaide

Canberra Hospital-Garran

Casey Hospital-Berwick

Christchurch Public Hospital-Addington

Coffs Harbour Health campus-Coffs Harbour

Concord Repatriation Hospital-Concord

Dandenong Hospital-Dandenong

Dubbo Base Hospital-Dubbo

Dunedin Public Hospital-Dunedin

Epworth Eastern Hospital-Box Hill

Epworth Hawthorn-Hawthorn

Epworth Hospital-Geelong

Epworth Hospital-Richmond

Fairfield District Hospital-Prairiewood

Fiona Stanley Hospital-Murdoch

Flinders Medical Centre-Bedford Park

Flinders Private Hospital-Bedford Park

Frankston Hospital-Frankston

Freemasons Hospital-East Melbourne

Fremantle Hospital-Fremantle

Friendly Society Private Hospital-Bundaberg West

Geelong Private Hospital-Geelong

Geelong Public Hospital-Geelong

Gold Coast Hospital Robina-Robina

Gold Coast Private Hospital-Parklands

Gold Coast Public Hospital-Southport

Gosford District Hospital-Gosford

Grace Hospital-Tauranga

Greenslopes Private Hospital-Greenslopes

Gretta Volum Day Surgery Centre-Geelong

Hastings Memorial Hospital-Camberley

Hawke's Bay Hospital-Camberley

Hobart Private Hospital-Hobart

Hollywood Private Hospital-Nedlands

8

Holmesglen Private Hospital-Moorabbin

Hornsby Ku-ring-gai Hospital-Hornsby

Innisfail Hospital-Innisfail

John Fawkner Hospital-Coburg

John Flynn Private Hospital-Tugun

John Hunter Hospital-New Lambton

Joondalup Health Campus-Joondelup

Kareena Private Hospital-Caringbah

Katherine District Hospital-Katherine

Knox Private Hospital-Wantirna

La Trobe Regional Hospital-Traralgon

Lake Macquarie Private Hospital-Gateshead

Launceston General Hospital-Launceston

Lingard Private Hospital-Merewether

Lismore Base Hospital-Lismore

Liverpool Hospital-Liverpool

Lyell McEwin Hospital-Elizabeth Vale

Macquarie University Hospital-North Ryde

Manly Hospital-Manly

Manukau Surgical Centre-Manurewa

Mater Adult Hospital-South Brisbane

Mater Hospital-Hyde Park-Townsville

Mater Hospital-Pimlico-Townsville

Mater Private Hospital-North Sydney

Melbourne Private Hospital-Parkville

Mercy Hospital-Epsom

Middlemore Hospital-Otahuhu

Monash Medical Centre-Clayton

Moorabbin Hospital-East Bentleigh

Mount Barker Hospital-Mt Barker

Mulgrave Private Hospital-Mulgrave

Nambour Selangor Private Hospital-Nambour

National Capital Private Hospital-Garran

Nelson Hospital-Nelson

New Bendigo Hospital-Bendigo

Newcastle Private Hospital-New Lambton Heights

Noarlunga Hospital-Noarlunga

Noosa Hospital-Noosaville

North Gosford Private Hospital-North Gosford

North Shore Private Hospital-St Leonards

North West Private Hospital-Burnie

North West Private Hospital-Everton Park

Northern Beaches Hospital-Frenchs Forest

Northern Hospital-Epping

Northpark Private Hospital-Bundoora

9

Norwest Private Hospital-Baulkham Hills

Norwest Private Hospital-Bella Vista

Ormiston Hospital-Botany Junction

Peninsula Private Hospital-Frankston

Perth Childrens Hospital-Nedlands

Pindara Private Hospital-Benowa

Port Macquarie Base Hospital-Port Macquarie

Port Macquarie Private Hospital-Port Macquarie

Prince of Wales Private Hospital-Randwick

Prince of Wales Public Hospital-Randwick

Princess Alexandra Hospital-Woolloongabba

Queen Elizabeth Hospital-Woodville West

Riverland Regional Hospital-Berri

Rosebud Hospital-Rosebud

Royal Adelaide Hospital-Adelaide

Royal Brisbane and Womens Hospital-Herston

Royal Darwin Hospital-Casuarina

Royal Hobart Hospital-Hobart

Royal Melbourne Hospital-Parkville

Royal North Shore Hospital-St Leonards

Royal Perth Hospital-Perth

Royal Prince Alfred Hospital-Camperdown

Royal Womens Hospital-Parkville

Sir Charles Gairdner Hospital-Nedlands

Southern Cross Hospital-Christchurch

Southern Cross Hospital-Wellington

Southern Highlands Private Hospital-Bowral

St Andrews Private Hospital-Adelaide

St Andrews Private Hospital-Ipswich

St Andrews Private Hospital-Toowoomba

St Andrews War Memorial Hospital-Brisbane

St George District Hospital-Kogarah

St George Private Hospital-Kogarah

St Georges Hospital-Christchurch

St JOG Hospital-Berwick

St JOG Hospital-Bunbury

St JOG Hospital-Geelong

St JOG Hospital-Midland

St JOG Hospital-Murdoch

St JOG Hospital-North Ballarat

St JOG Hospital-Subiaco

St John's Hospital-South Hobart

St Vincents Private Hospital-Darlinghurst

St Vincents Private Hospital-East Lismore

St Vincents Private Hospital-Fitzroy

10

St Vincents Private Hospital-Launceston

St Vincents Private Hospital-Northside

St Vincents Private Hospital-Werribee

St Vincents Public Hospital-Darlinghurst

St Vincents Public Hospital-Fitzroy

Steele Street Clinic-Devonport

Stirling Hospital-Stirling

Strathfield Private Hospital-Strathfield

Sunshine Coast Private Hospital-Buderim

Sunshine Coast University Private Hospital-Birtinya

Sunshine Coast University Public Hospital-Birtinya

Sunshine Hospital-St Albans

Sutherland District Hospital-Caringbah

Sydney Adventist Hospital-Wahroonga

Sydney South West Private Hospital-Liverpool

Tamworth Base Hospital-Tamworth

Taranaki Base Hospital-Westown

Tauranga Public Hospital-Tauranga

The Mount Hospital-Perth

The Nepean Hospital-Penrith

The Nepean Private Hospital-Kingswood

The Prince Charles Hospital-Chermside

The Surgery Centre-Hurstville

The Tweed Hospital-Tweed Heads

The Vein Centre-Richmond

The Wesley Hospital-Auchenflower

Toowoomba Base Hospital-Toowoomba

Townsville Hospital-Townsville

Varsity Lakes Day Hospital-Varsity Lakes

Vascular Solutions-Subiaco

VCCC-Parkville

WA Vascular Centre-Bassendean

Wagga Wagga Base Hospital-Wagga Wagga

Wagner rooms-Melbourne

Waikato Hospital-Hamilton

Warringal Private Hospital-Heidelberg

Wauchope District Hospital-Wauchope

Waverly Private Hospital-Mt Waverly

Wellington Hospital-Wellington

Western Hospital-Footscray

Western Private Hospital-Footscray

Westmead Hospital-Westmead

Westmead Private Hospital-Westmead

Williamstown Hospital-Williamstown

Wimmera Base Hospital-Horsham

11

Wollongong Hospital-Wollongong

Wollongong Private Hospital-Wollongong

Wyong Public Hospital-Kanwal

The mean number of operations per hospital was 216 with a range of 1-1,491

The distribution of procedures by patient type is shown in Fig. 3. The majority were arterial patients followed by venous disease then renal disease.

Fig 3. Patient type 2019

The distributions of procedures in the arterial category are shown in Fig. 4. The majority were for chronic limb operations followed by aneurysms then acute limb procedures.

12

Fig 4. Arterial categories 2019 (n=23,501)

In the 23,501 arterial operations the risk factors present are shown in Fig. 5. Hypertension was the most frequent risk factor recorded followed by ischaemic heart disease (IHD) then diabetes.

Fig 5. Risk factors in arterial operations 2019 (Creatinine = >150mMol/L, Smoking = current)

13

Aortic Surgery There were 2,855 Aortic (discharged) procedures performed in 2019. This category includes aneurysmal disease (emergency and elective), open and endoluminal (ELG) procedures and aortic operations for non-aneurysmal disease.

Age and gender are shown in Fig. 6

The distribution of procedures and crude mortality is shown in Table 1.

Table 1. Aortic surgery raw data

Category

Total

Mortality (%)

All Aortic procedures 2853 5.2 Open Aortic surgery 999 10.9 Open AAA 628 9.9 Open AAA-elective 396 4.0 Open AAA-ruptured 137 29.2 AAA-EVAR-elective 1352 0.7 AAA-EVAR-ruptured 80 18.8 Non-aneurysm abdominal aortic surgery 349 12 Thoracic ELG 254 4.7 Open Thoracoabdominal 10 30

14

i) Open aortic surgery

This includes all aneurysm and non-aneurysm surgery. 193 surgeons performed an average of 5 procedures. The indications for the 349 non-AAA procedures are shown in Table 2.

Table 2. Non-aneurysm open aortic surgery

Indication Total Died

Claudication 96 4

Mesenteric ischemia 55 19

Rest pain 55 1

Acute ischemia 50 10

Ulcer/gangrene(arterial) 24 0

Trauma(iatrogenic)-haemorrhage 19 0

Neoplasm-malignant 9 0

Aortoenteric fistula-secondary 7 3

Dissection 7 1

Trauma(non iatrogenic)-haemorrhage 6 2

Bypass / Stent graft / Patch sepsis 4 1

Endoleak 4 0

Infection 3 0

Retrieval device/FB 3 0

Renal a stenosis/refractory hypertension 2 0

Trauma(non iatrogenic)-occlusion 2 1

Aortoenteric fistula-primary 1 0

AV Fistula closure 1 0

Outcomes for Open Aortic Surgery

This data was risk-adjusted using predictive models obtained by logistic regression analysis (see Appendix 2-statistical methods). A multilevel model was not significant so standard binary logistic regression analysis was used.

The open aortic surgery model displayed excellent calibration (a measure of the ability to predict mortality across the spectrum of low and high risk patients), determined by “goodness of fit” tests that do not show a difference, as well as good discrimination (the ability of the model to predict mortality in any particular patient) as determined by the area under the ROC, with a value of this C-statistic of > 0.7 signifying good discrimination.

The ROC graph for the model for open aortic surgery is shown in Fig. 7 with a C-statistic of 0.83.

15

Fig 7. ROC for mortality after open aortic surgery model.

Table 3 shows the significant variables used in the model for all open aortic surgery 2019.

Parameter Odds Ratio P (>|Z|) Aortoenteric Fistula 7.760912 P = 0.008 Ruptured AAA 2.695719 P = 0.0016 Mesenteric ischemia 6.027512 P < 0.0001 ASA Status(4) 3.897836 P < 0.0001 ASA Status(5) 4.774147 P = 0.0001 71-80 years 2.629658 P = 0.0002 81-90 years 4.191389 P < 0.0001 Male 0.433385 P = 0.0007

0.0

00.2

50.5

00.7

51.0

0

Sen

sitiv

ity

0.00 0.25 0.50 0.75 1.001 - Specificity

Area under ROC curve = 0.8271

16

Once a predictive model is obtained, probabilities of mortality are obtained from the model and used to display risk-adjusted mortality based upon an expected mortality rate for each patient.

Funnel plots have been constructed and were plotted by including 59 consultants where 6 or more cases were performed during 2019. This plot shows adjusted standardized mortality rate on the Y-axis against total cases done on the X-axis. Another graph using 95% and 99% Poisson confidence intervals of the expected mortality for each surgeon is superimposed. This produces an easy to read graph showing any outliers. The mortality rate was 10.9% for open aortic surgery.

17

Fig 8. Risk-adjusted funnel plot for open aortic surgery for consultants with 6 or more cases (52)

Outliers

No outliers were identified.

Open AAA

628 patients underwent surgery for open AAA in 2019. This dataset was restricted to patients with abdominal aneurysm repair, excluding thoraco-abdominal aneurysms. This allowed comparison of postoperative complications between 491 intact (elective, mycotic, painful, occluded) aneurysms and 137 ruptured AAA (Table 4). Mean aneurysm diameter was 66mm.

Table 4. Complications after intact and ruptured AAA repair

Complication Intact AAA (491) Ruptured AAA (137)

AMI 9(1.8%) 7(5.1%) Gut ischaemia 11(2.2%) 9(6.6%) Renal failure/impairment 34(6.9%) 31(22.6%) Died 22(4.5%) 40(29.2%)

0

20

40

60

Mort

alit

y %

5 10 15 20 25Total

Consultant Sign. 5% Sign. .2%

Mortality after open aortic surgery 2019

10.9

18

Outcomes

Predictive variables for the model are shown in table 5. Excellent discrimination was obtained with a c-statistic of 0.84. A multilevel model was not used as it was not significantly different from the binary logistic regression model.

Table 5. Significant variables in the Open AAA model 2019.

Parameter Odds Ratio P (>|Z|) >4 L bloodloss 3.18536 P = 0.004 Ruptured 8.121692 P < 0.0001 IHD 2.211692 P = 0.0093 Age>80 2.611594 P = 0.0082 Male 0.32061 P = 0.0007

19

Fig 9. Risk-adjusted funnel plot for open AAA repair where surgeons performed >6 cases (25)

Outliers: There were no outliers for open AAA surgery. Raw mortality was 9.9%

iii) Endoluminal abdominal aortic surgery

Abdominal aortic aneurysm

1,614 non-thoracic ELG were inserted during 2019. 87% patients had percutaneous access with closure device. Mean aneurysm diameter was 58mm. There were 31 type 1, 43 type 2 and 7 type 3 endoleaks. There were 13 occluded limbs and 2 conversions to an open repair. GA was used in 94.5%.

The indication for EVAR was not confined to AAA as shown in Table 6.

0

10

20

30

40

50

Mort

alit

y %

5 10 15 20Total

Consultant Sign. 5% Sign. .2%

Mortality after open AAA 2019

20

Table 6. Indications for EVAR 2019

Indication Total

Aneurysm-elective 1352

Aneurysm-pain 115

Aneurysm-ruptured 80

Aneurysm-mycotic 18

Endoleak 17

Aneurysm-occluded 10

Dissection 10

Aortoenteric fistula-secondary 4

Aortoenteric fistula-primary 3

Claudication 3

Acute ischemia 1

Aneurysm-false(non iatrogenic trauma) 1

Comparison of complications between intact and ruptured ELG insertion is shown in Table 7 (the intact group includes AAA and other ELG inserted for non-AAA).

Table 7. Complications after intraabdominal ELG (n = 1,614)

Complication Intact Aorta (1,534) Ruptured AAA (80)

Conversion 2 0 AMI 11 1

Gut ischaemia 2 1 Renal failure/impairment 24 8 Endoleak type 1 30 1

Endoleak type 2 42 1

Endoleak type 3 6 1 Died 12(0.8%) 15(18.8%)

21

The types of device used for ELG is shown in table 8.

Device Endurant

Zenith Alpha

Excluder

Cook low profile

Zenith Fenestrated

Cook low profile with spiral limb(s)

Cook with side branches

Endologix

Zenith Flex(non-fenestrated)

Other hybrid combination

Cordis Incraft

Anaconda(non-fenestrated)

Zenith T Branch

Zenith body with Gore limb(s)

Trivascular Ovation (Prime)

Aorfix

Zenith body with Anaconda limb(s)

Zenith body with Endurant limb(s)

Ancure

Cook low profile, Zenith Alpha

Cook with side branches, Cook low profile

Cook with side branches, Endurant

Cordis Incraft, Anaconda(non-fenestrated)

Endologix, Excluder

Endologix, Other hybrid combination

Excluder, Zenith Fenestrated

Excluder, Zenith Alpha

Nellix

Zenith Fenestrated, Cook with side branches

iv) Fenestrated and branched ELG

The configuration of all ELG is shown in Table 9. The subsets of branched and fenestrated grafts are evident; 11.7% were fenestrated with a mortality of 7/190 (3.7%) vs non-fenestrated 20/1,424 (1.4%) P=0.03. Endoleaks occurred in 5.2% of non-fenestrated vs 3.7% in fenestrated ELG (ns).

22

Table 9. Configuration of ELG 2019

Configuration Total

Bifurcated 1268

Fenestrated Renal(s)-SMA-Coeliac 92

Tube 71

Fenestrated Renal(s)-SMA 45

Bifurcated-bifurcated(+/- IBD) 36

Fenestrated + Branched endograft 24

Aorto-uni-Iliac and Fem fem bypass 22

Fenestrated both Renals 21

Branched endograft R Iliac 12

Branched endograft L Iliac 9

Aorto-uni-iliac-no x-over 8

Fenestrated L Renal 3

Fenestrated R Renal 2

Fenestrated SMA-Coeliac 1

Outcomes

Mean mortality for all EVAR (for AAA only) was 1.6%. The c-statistic was 0.86. Significant variables in the model were gender, Fenestrated graft and ruptured AAA.

Table 10. Significant variables for mortality after EVAR 2019

Parameter Odds Ratio P (>|Z|) Fenestrated 9.062313 P = 0.0006 Ruptured 63.751574 P < 0.0001 Male 0.381449 P = 0.0515

23

iv) Thoracic and thoraco-abdominal procedures

Endoluminal. Of the thoracic and thoracoabdominal ELG (n=254), the group consisted of dissecting aneurysms (23), non-dissecting aneurysms (86), acute dissection (44), chronic dissection (39), traumatic aortic tear (34) and penetrating ulcer (28). There were 12 deaths (4.7%), not considered significant for this procedure. 111 surgeons inserted a mean of 2 ELG with a range from 1-12. Configuration is shown in Table 11.

Configuration Total

Overlapping Stent grafts 126

Single Stent graft 120

Stent graft(s) with distal bare stent 7

Stent graft(s) with intra-abdominal fenestration(s) 1

The following devices were inserted in patients having stents/stent grafts in the thoracic aorta (Table 12).

Device Zenith Alpha

Medtronic

Gore C-TAG

Zenith TX2

Endospan Nexus

There were 9 patients with paraplegia (3.5%) and 6 strokes (2.4%) following TEVAR. 8 patients had renal failure or impairment and 1 developed intestinal infarction. There was 1 type 1, and 1 type 3 endoleaks. No patients required conversion to open. Breakdown of complications by aetiology is shown in Table 12.

24

Table 13. Complications according to the main pathology types (n=254)

Pathology Total Mortality Stroke Paraplegia Aneurysm(dissecting) 23 1 1 Aneurysm(non-dissecting) 86 3 3 4 Dissection-acute 44 3 2 Dissection-chronic 39 1 1 Traumatic tear 34 2 1 Penetrating ulcer 28 2 2 1

Outcomes

No predictive model was produced. Because of low numbers no outlier detection could be run for TEVAR in 2019. A cumulative report will be produced in 2021 to correct this, as was the case in the 2015-2017 report.

Open. There were 10 open thoracoabdominal procedures with 3 deaths They were performed by 9 surgeons and one surgeon had performed 2 cases. There was a single stroke and no paraplegia was recorded in this cohort. There were 2 aneurysm ruptures with one death and 5 intact aneurysms with a single mortality in a complex type 4 on bypass, with uncontrollable bleeding after flow restoration. This was not considered significant by the audit monitoring committee for this procedure.There were 3 dissections with a single mortality. Mean diameter of the aneurysms was 64mm.

Carotid Surgery

There were 2,061 carotid interventions, 1,905 carotid endarterectomies (CEA) and 156 carotid stents

(CAS).

i) Carotid Endarterectomy

The indications for CEA are shown in Fig.12 with 23% having no symptoms.

25

Fig 12. Indication for CEA

The time from onset of symptoms to surgery in symptomatic patients was < 48 hours in 1%, < 2

weeks in 59%, 2-4 weeks in 20% and > 4 weeks in 20%. NICE guidelines recommend that the goal

should be to operate within 2 weeks from the onset of symptoms to have the lowest stroke

incidence. General anaesthesia was used in 80% of the patients.

Eversion endarterectomy was performed in 13.4% of patients and 43% were shunted. Patches were

used in 87% of CEA (Table 14).

Table 14. Patches after CEA.

Patch Total

Polyurethane 745

Pericardium 537

Dacron 211

No patch/conduit 167

PTFE 76

Prosthetic (Other) 55

GSV-reversed 19

Neck vein 8

Homograft 4

Ext carotid 2

GSV-non reversed 1

Vein (Other) 1

26

Complications after CEA are shown in table 15.

Table 15. Complications after CEA (n= 1,905)

Complication Percent

Haemorrhage requiring exploration 2.6

Cranial nerve trauma 0.7

Myocardial infarction 0.6

Major/minor stroke 0.9

TIA 0.4

Hyperperfusion 0.3

Death 0.3

Stroke or death 1.1

Outcomes

Only 1 variable was significant in the model for stroke/death (Table 16). Thus a non-risk-adjusted

funnel plot was constructed.

Table 16. Significant variables for S/D after CEA 2019

Parameter Odds Ratio P value ASA4 4.863636 P = 0.0133

Because the numbers were low, only those surgeons (85) who performed 10 or more CEA were

assessed by a funnel plot. The mean stroke/death(S/D) rate was 1.1% and no outliers were

apparent. Symptomatic S/D rate was 1.1% and Asymptomatic S/D was 0.9%. postop S/D rate for

stroke as the indication for operation was 1.6%.

27

Fig 13. Non risk-adjusted Funnel plot for stroke and death after CEA 2019

0

5

10

15

20

S/D

%

10 15 20 25 30 35Total

Consultant Sign. 5% Sign. .2%

Stroke/death after CEA 2019

1.1

28

ii) Carotid Stents

156 carotid stents were performed in 2019.

Indications for CAS are shown in Fig 15, with the most frequent being asymptomatic, then TIA.

Technical details. n=156

Access was via a long sheath in 111 and via a short sheath with guiding catheter in 45. There was a

type 1 arch in 93, type 2 in 58 and type 3 in 5 patients.

Cerebral protection devices used are shown in table 17. No protection device was employed in 17

patients.

Filter

Emboshield

Nav 6

None

Filterwire EX

Angioguard

SpiderFX

Accunet

Neuroshield

29

Stent types are shown in table 18.

Stent

Xact

Covered stent

Precise

Wallstent

Casper

CGuard

Angioplasty only

Tapered

ProtegeRX

Medtronic Cristallo

Smart

Outcomes

There was a single post op stroke and 1 death giving a stroke and death rate of 2/156(1.3%). Both

patients were symptomatic. There were no AMIs and 1 had renal impairment.

Infrainguinal bypass

1,693 Infrainguinal bypasses (IIB) were performed in 2019. The average age of patients was 68 with

the M: F ratio of 3.8:1. General anaesthetic was used in 96%.

Indications for surgery are shown in Fig 16 with tissue loss being the most frequent.

30

31

Fig. 17 Conduits for infrainguinal bypass.

Bypass configuration is shown in Fig 18.

32

Post-operative complications are shown in table 19 (n = 1,693)

Complication Percent

Myocardial infarction 1.4 Stroke 0.4 Renal impairment/ failure 1 Wound complications 6 Haemorrhage requiring reoperation 3.5 Death 0.8

Outcomes

i) Occlusion

A multilevel logistic regression model for occlusion after IIB was obtained. Variables included are

shown in table 20.

Parameter Odds Ratio P (>|Z|) 1Vessel 1.972003 P = 0.0037 Composite 3.980883 P = 0.0006 Male 0.455132 P = 0.0012

Occlusion rates were assessed using a risk adjusted funnel plot for those consultants that performed

10 or more bypasses (Fig 19). No outliers were detected for 2019. The mean occlusion rate was 4.8%

and mortality was 0.8%.

33

Fig 19. Risk adjusted funnel plot for occlusion after IIB 2019 (> 9 cases)

Popliteal Aneurysm: There were 230 bypasses for aneurysm (elective, occluded, pain or rupture).

The graft occlusion rate for these was 2.6% and the major amputation rate was 0%. In non-aneurysm

patients the graft occlusion rate was 5.1% and the amputation rate was 1.3 %.

Claudicants vs tissue loss: In the 386 claudicants, the occlusion rate was 2.8% and there was 1

amputation. In 492 patients with tissue loss the occlusion rate was 5.5% and the amputation rate

was 1.6%.

ii) Amputation

The limb salvage rate was 98.9%. 19 limbs were amputated and 5 of these occurred with a patent

graft. 3 patients in this subgroup were diabetic.

0

10

20

30

Occlu

sio

n %

10 15 20 25 30Total

Consultant Sign. 5% Sign. .2%

Occlusion after IIB 2019

4.8

34

Arteriovenous Fistulae

2,653 patients had an arteriovenous fistula (AVF) placed in 2019. The locations of AVF are shown in

Fig 20.

Fig 20. AVF configuration

The majority of AVF were autogenous (2,481) and 6.5% were prosthetic. The conduits used are

shown in Fig 21.

35

Fig 21. Conduits used

Outcomes

There were 37 occlusions (1.6%). Autogenous fistulae occluded in 34/2481(1.4%) and prosthetic

fistulae occluded in 3/172(1.7%). 6 patients had a steal syndrome, 1 of these was a thigh loop and 2

occurred in a wrist fistula. The other 3 were in brachiocephalic AVF.

No model was obtained for occlusion after AV Fistula in 2019. The only significant variable was the

use of an Omniflow graft with 2 occlusions out of 14 cases, both by different surgeons.

36

Data validation and conclusions

This audit report has been the culmination of much hard work by the committee and the

contributing membership. The most important conclusion is that the standard of Australasian

vascular surgery remains high with excellent outcomes in all the selected areas of audit. The

outcomes chosen for audit in these 4 procedures are the best method of assessing the clinical and

technical skill of a vascular surgeon. The most important facet of an activity such as this remains the

“audit of the audit”, and there are methods that were established during the inaugural year for both

external and internal validation of this activity. External validation for Australian data has compared

data capture between the AIHW database and the AVA (by financial year for the preceding years as

data becomes available). Overall capture in the AVA for all Australian private and public hospital

operations in the 4 index procedures has been shown to be 63% compared to AIHW data up to the

2017/8 financial year. Data validation in the private sector only is available by accessing Medicare

data. This is available for all billed procedures, which excludes VA and public patients. This data has

been analysed for calendar years 2010-9 for the following categories of patient (Australia only):

Carotid endarterectomy

Item numbers 33500 and 32703

Intact AAA (open and endoluminal)

Item numbers 33112, 33115, 33116, 33118, 33119, 33121, 33136, 33139

Infrainguinal bypass

Item numbers 32739, 32742, 32745, 32748, 32751, 32754, 32757, 32763, 33050, 33055

AV Fistula

Item numbers 34503, 34509, 34512

This data was compared with AVA data over the same period after exclusion of public and VA

patients. This shows that there is poor entry of private data, but is just over 50% again but has

dropped from last year. Further measures are required to increase this percentage.

Fig 23. Private practice participation in the AVA for Australia 2010-2019

37

Internal validation was performed at the end of 2017 comparing a 5% sample of patients with the

actual case notes by nominated members at each hospital. This showed that data entry was of high

quality with only 2.7% having incorrect field data entered out of a total of 3,225 fields studied. This

study is repeated every 3 years. Performance of vascular surgery in Australasia is at a high standard

and our Society is enhanced by the existence of the AVA, especially with its unique audit loop.

Members can continue to participate in the knowledge that it is a completely confidential activity,

monitored by a committee that has a dual role of scrutiny of outcomes together with a genuine

concern for the natural justice of members.

C Barry Beiles, Administrator

38

Appendix 1

Algorithm for audit

Note 1. The members of the ANZSVS Audit Monitoring Committee (AMC) are responsible for determining the thresholds for complications warranting review, after discussion and agreement by the members. Where appropriate, the thresholds used by the ACHS may be the limit chosen. Note 2. If it is not possible for the independent reviewer chosen by the member and the AMC to reach consensus, the issue will be referred to the Board of Vascular Surgery for a final determination of satisfactory or unsatisfactory performance or other recommendation. Note 3. The algorithm does not envisage advice to stop all operating unless audit showed unsatisfactory results in all types of operations performed. Thus the surgeon would only cease performing that particular operation that gave unsatisfactory results. Referral to the Medical Board may result in the suspension of all operating rights.

If there are continuing issues with the surgeon performing operations at an unsafe level then notification of the concerns of the AMC may be made to the Medical Board after discussion in writing with the president of the ANZSVS.

AMC notifies the member.

The member & AMC review the relevant

cases. Data relevant to the review are

collated & checked for accuracy.

The review demonstrates satisfactory

results.

Continue audit.

The member may appoint a surgical

colleague to assist with review.

The surgeon may request review of the

data by a member of the ANZSVS.

The review demonstrates unsatisfactory

results. De-identified data sent to the

President of the ANZSVS for review. The

surgeon is informed in writing of the

outcome of the review and advised to

stop performing the procedure(s).

Surgeon does not want to stop

performing the procedure(s) or does not

respond to original data request.

Chair of the ANZSVS notifies: Head of the appropriate vascular unit. Director of Medical Services at the hospital. RACS via President of ANZSVS

Surgeon requests temporary halt,

pending re-training and re-

credentialing by Board of Vascular

Surgery.

The surgeon agrees to stop the

procedure on a permanent basis.

AUDIT MONITORING COMMITTEE (AMC) IDENTIFIES A "VARIANT RESULT"

39

Appendix 2

Statistical methods

When performing institutional or individual comparisons for outcomes of health data, it is important

to recognise that this has been fraught with difficulties in the past. The now discredited league tables

are misleading and have been replaced with funnel plots, which are easy to interpret at a glance. The

league table approach has been used to rank institutions based on performance, and this has led to

"gaming", whereby institutions tackling the more complicated high-risk cases have avoided these

procedures in order to improve their position in the table. There is also a 5% risk that a hospital or

surgeon will be at the bottom of the table by chance, as these tables use 95% confidence intervals. It

should also be recognised that it is a statistical certainty that an institution or surgeon can have a run

of bad luck, and while they might reside at the bottom of the table in 1 year, this may be an isolated

phenomenon.

Whichever method is used in assessing performance, some method of risk-adjustment is important,

so that those hospitals or surgeons undertaking the high-risk cases will not be disadvantaged. It is

recognised that methods of obtaining risk-adjustment are not an exact science, but the most widely

utilised technique applied to outcomes that are ‘binary’ (where the outcome is one of 2 choices, ie.

death or survival; patency or occlusion), is multilevel logistic regression analysis. Multilevel analysis

determines the effect of the hospital on patients treated by the same surgeon at different locations.

The outcome variable is called the dependent variable, and the variables that significantly affect the

outcome are called the independent variables. These variables are accepted if the P value is < 0.05.

An acceptable model is then produced that aims to provide good predictive qualities (called

"discrimination") and this predictive ability should persist for cases with both low and high risk of an

adverse outcome (called "calibration"). We have been able to produce good models for mortality

following open aortic, open aneurysm, EVAR, stroke/death after carotid endarterectomy, occlusion

after lower limb bypass and occlusion after AVF creation. The link test was run after each logistic

regression to confirm that the model was correctly specified.

Once a model has been established, it will provide an expected risk of an adverse outcome for each

patient in the population studied, based on the presence or absence of the statistically significant

variables identified by the logistic regression procedure. This is then applied in the methods chosen

to display the data. Statistical analysis was performed using Stata version 13.1 (Statacorp.

4905 Lakeway Drive College Station, Texas 77845 USA) and StatsDirect statistical software (England:

StatsDirect Ltd. 2008)

Data display

Funnel plots have been adapted from a technique used to establish publication bias in meta-

analyses. The adverse event rate is plotted on the Y axis, with the total number of cases on the X axis

40

and Poisson 95% and 99% confidence intervals using the pooled adverse event rate for the whole

group superimposed on the scatter plot. The data is risk adjusted (where a robust predictive model

has been obtained) by plotting the adverse event rate as a standardised mortality/event ratio

(Observed/ Expected rate x overall event rate expressed as a percentage). The expected rate for

each patient is derived from the logistic regression analysis. Non risk-adjusted funnel plots are

displayed using the percent adverse event on the Y-axis and using a binomial distribution. These

plots were obtained by using the funnelcompar module in Stata. The graph is easily interpreted

because any consultant falling outside the upper 95% confidence interval should be scrutinised to

see if there is a problem in processes, using careful clinical appraisal. Conversely, consultants falling

below the lower 95% confidence interval are performing much better than the majority.

41

Appendix 3

Features of the AVA application

This is a web-based database in SQL residing on a secure server (Microsoft Azure) within Australia and is compatible with all browser platforms. Data capture is exclusively via the web portal. A mobile-friendly modification has recently been designed.

1. Security and performance:

a) Uptime – Application and database up-time is greater than 99%

b) Backup Services - Daily database and application backup

c) Security services - Enterprise Firewalls, Intrusion Prevention Systems, and Anti-Virus Protection

d) Disaster recovery - Daily backups featuring file recovery, data de-duplication, redundant block elimination, over the wire encryption and offsite storage of backup data

e) Logon is only permitted by Surgeon code and password

f) The ability to view reports is determined by the status of the user. Full members of the ANZSVS have the ability to view all reports, and there is the ability to view the user’s outcomes in the 4 categories of audit in real time compared to the peer group. There is also a category of data manager for a unit or hospital (e.g. vascular trainee) that is granted access to enter data for the surgeons who work in their unit. They have no access to the private patient data for those surgeons.

2. Scalability:

The application is capable of handling 200 simultaneous users

3. Role based data updates:

Modification of data entered in the discharge/complication form fields after user logoff is only allowed by the administrator. Addition of data is allowed by all users. Deletion of records is only allowed by the administrator.

4. Privacy and confidentiality:

Compliance with privacy legislation is current and patient identifiers are encrypted and the database is securely stored by the Server. Confidentiality of patient details is thus assured. Ethics committee approval has been obtained for this activity. Confidentiality of member’s identity is assured by the storage of the surgeon code with legal representatives of the ANZSVS. The only situation where the identity of a surgeon would be allowed is in the event of the examination of the member by the AMC after possible underperformance has been identified by the statistical analysis. Commonwealth legislation identifying the AVA as a privileged quality assurance activity has been obtained in both Australia and New Zealand. Any identification of participating members outside of the strict algorithm of the audit process is punishable by a significant financial penalty and a maximum 2 year custodial sentence. An important feature of the AVA is the independence provided by total ownership of the data. This has been possible because the ANZSVS has self-funded the establishment and maintenance

42

costs.

5. Data reliability:

Strict data validation criteria prevent erroneous data entry and there is no ability for free text data entry, except for 2 “comment” boxes in the operation and discharge forms. Drop down menus allow choices to appear that are based upon selections made in previous fields. This diminishes the ability to enter incorrect data.

6. Flexibility:

The application has been designed to allow alterations to the menu choices by the administrator. This has ensured that unusual operations can be entered. The application captures all endovascular procedures where appropriate and the vascular surgical trainees extract data from the AVA to submit their logbooks to the Board of Vascular Surgery.

7. Benefits for the user: The ability to compare real time outcomes by surgeon and /or hospital with the membership as a whole is very attractive. Also, there is the ability to conduct unit or personal audit using the reports specifically designed for this purpose. There is the ability to export data extracts, which represent a spreadsheet containing every field for each patient. This allows filtering to manipulate data in any form the user requires for any purpose. Logbook reports are also available for trainees and members. Participation in the AVA has been approved as a recognised audit activity by the Royal Australasian College of Surgeons for the purpose of re-accreditation. Participation also allows the user access to de-identified data for the purpose of research or in the event of an inquiry into one’s performance by a hospital or medico legal proceeding. A certificate of participation is issued annually upon application. This certificate is mandatory for retention of membership of the Society since 2019.

References

1) Spiegelhalter D. Funnel plots for comparing institutional performance. Stat Med. 2005 Apr 30;

24(8):1185-202.

3) Bourke BM, Beiles CB, Thomson IA, Grigg MJ, Fitridge R. Development of the Australasian Vascular

Surgical Audit. J Vasc Surg 2012; 55:164-70

4) Beiles CB, Bourke B, Thomson I. Results from the Australasian Vascular Surgical Audit: the

inaugural year. ANZ Journal of Surgery. 2012; 82: 105-111

5) Sanagou M, Wolf R, Forbes A, Reid C. Hospital-level associations with 30-day patient mortality

after cardiac surgery: a tutorial on the application and interpretation of marginal and multilevel

logistic regression. BMC Medical Research Methodology 2012; 12:28.

http://www.biomedcentral.com/1471-2288/12/28

6) Beiles C B and Bourke B M. Validation of Australian data in the Australasian Vascular Audit. ANZ

Journal of Surgery. 2014; 84: 624-627